TiSSSKJr SSKJrJrJrJrJr SSKJ r SSK J r SSK J r SSKJr SSKJrJrJrJrJrJrJrJrJrJrJrJrJrJrJrJ r SSK!J"r"J#r# "S S \5r$g ) ) annotations)OptionalSequenceTupleTypeVarUnion) GraphProto) get_schema) TypeAlias)Opset16)BFLOAT16BOOL COMPLEX64 COMPLEX128DOUBLEFLOATFLOAT16INT8INT16INT32INT64STRINGUINT8UINT16UINT32UINT64)OpOpsetcj\rSrSr%Sr\"S\\5r\ \ \ \ \ \\\\\\\\4 rS\S'SSS.S(Sjjr\"S \ \ \ \ 5r\"S \\5rS)SS S S .S*Sjjjr\"S\\5r\ \ \ \ \ \\\\\\\\4 rS\S'SSS.S+Sjjr\"S\\5r\ \ \ \ \ \\\\\\\\4 rS\S'SSS.S,Sjjr\"S\ \ \ \ 5r \ \ \ 4r!S\S'S)SSSS.S-Sjjjr"\"S\\5r#\"S\ \ \ \ 5r$\ \ \ \ \ \\\\\\\\4 r%S\S'SS.S.Sjjr&\"S \ \ \ \ 5r'\"S!\\5r(S/SS".S0S#jjjr)\"S$\*\+\*\,\*\-\*\ \*\ \*\ \*\\*\\*\\*\\*\.\*\\*\\*\\*\5r/\"/S%P\*\+P\*\,P\*\-P\*\ P\*\ P\*\ P\*\P\*\P\*\P\*\P\*\.P\*\P\*\P\*\P\*\P\+P\,P\-P\ P\ P\ P\P\P\P\P\.P\P\P\P\P76r0S1S&jr1S'r2g )2Opset17+c2[R"USS5$)N)r__new__)clss b/var/www/html/ai-image-ml/venv/lib/python3.13/site-packages/onnxscript/onnx_opset/_impl/opset17.pyr%Opset17.__new__,s}}S"b))T1_BlackmanWindowr T2_BlackmanWindowoutput_datatypeperiodiccf[SSS5n[USU5nU"URXA5UUS.6$)u[🌐 BlackmanWindow(17)](https://onnx.ai/onnx/operators/onnx__BlackmanWindow.html#blackmanwindow-17 "Online Documentation") Generates a Blackman window as described in the paper https://ieeexplore.ieee.org/document/1455106. Args: size: (non-differentiable) A scalar value indicating the length of the window. output_datatype: The data type of the output tensor. Strictly must be one of the values from DataType enum in TensorProto whose values correspond to T2. The default value is 1 = FLOAT. periodic: If 1, returns a window to be used as periodic function. If 0, return a symmetric window. When 'periodic' is specified, hann computes a window of length size + 1 and returns the first size points. The default value is 1. BlackmanWindowr$r#r-r r_prepare_inputsselfsizer.r/schemaops r'r1Opset17.BlackmanWindow@sC.,b"5 & /  ! !& /+  r)T1_DFTT2_DFTNraxisinverseonesidedcj[SSS5n[USU5nU"URXaU5UUUS.6$)u[🌐 DFT(17)](https://onnx.ai/onnx/operators/onnx__DFT.html#dft-17 "Online Documentation") Computes the discrete Fourier transform of input. Args: input: (non-differentiable) For real input, the following shape is expected: [batch_idx][signal_dim1][signal_dim2]...[signal_dimN][1]. For complex input, the following shape is expected: [batch_idx][signal_dim1][signal_dim2]...[signal_dimN][2]. The first dimension is the batch dimension. The following N dimensions correspond to the signal's dimensions. The final dimension represents the real and imaginary parts of the value in that order. dft_length: (optional, non-differentiable) The length of the signal as a scalar. If greater than the axis dimension, the signal will be zero-padded up to dft_length. If less than the axis dimension, only the first dft_length values will be used as the signal. It's an optional value. axis: The axis on which to perform the DFT. By default this value is set to 1, which corresponds to the first dimension after the batch index. Negative value means counting dimensions from the back. Accepted range is $[-r, -2] \cup [0, r-2]$ where `r = rank(input)`. The last dimension is for representing complex numbers and thus is an invalid axis. inverse: Whether to perform the inverse discrete fourier transform. By default this value is set to 0, which corresponds to false. onesided: If onesided is 1, only values for w in [0, 1, 2, ..., floor(n_fft/2) + 1] are returned because the real-to-complex Fourier transform satisfies the conjugate symmetry, i.e., X[m, w] = X[m, n_fft-w]*. Note if the input or window tensors are complex, then onesided output is not possible. Enabling onesided with real inputs performs a Real-valued fast Fourier transform (RFFT). When invoked with real or complex valued input, the default value is 0. Values can be 0 or 1. DFTr$r#r<r2)r5input dft_lengthr=r>r?r7r8s r'rA Opset17.DFTcsG^E2r* eV $  ! !& <   r)T1_HammingWindowT2_HammingWindowcf[SSS5n[USU5nU"URXA5UUS.6$)u[🌐 HammingWindow(17)](https://onnx.ai/onnx/operators/onnx__HammingWindow.html#hammingwindow-17 "Online Documentation") Generates a Hamming window as described in the paper https://ieeexplore.ieee.org/document/1455106. Args: size: (non-differentiable) A scalar value indicating the length of the window. output_datatype: The data type of the output tensor. Strictly must be one of the values from DataType enum in TensorProto whose values correspond to T2. The default value is 1 = FLOAT. periodic: If 1, returns a window to be used as periodic function. If 0, return a symmetric window. When 'periodic' is specified, hann computes a window of length size + 1 and returns the first size points. The default value is 1. HammingWindowr$r#r-r2r4s r'rHOpset17.HammingWindowsA.OR4 ov .  ! !& /+  r) T1_HannWindow T2_HannWindowcf[SSS5n[USU5nU"URXA5UUS.6$)u[🌐 HannWindow(17)](https://onnx.ai/onnx/operators/onnx__HannWindow.html#hannwindow-17 "Online Documentation") Generates a Hann window as described in the paper https://ieeexplore.ieee.org/document/1455106. Args: size: (non-differentiable) A scalar value indicating the length of the window. output_datatype: The data type of the output tensor. Strictly must be one of the values from DataType enum in TensorProto whose values correspond to T2. The default value is 1 = FLOAT. periodic: If 1, returns a window to be used as periodic function. If 0, return a symmetric window. When 'periodic' is specified, hann computes a window of length size + 1 and returns the first size points. The default value is 1. HannWindowr$r#r-r2r4s r'rMOpset17.HannWindowsA.L"b1 lF +  ! !& /+  r)T_LayerNormalizationU_LayerNormalizationg>r=epsilon stash_typecj[SSS5n[USU5nU"URXqX#5UUUS.6$)u [🌐 LayerNormalization(17)](https://onnx.ai/onnx/operators/onnx__LayerNormalization.html#layernormalization-17 "Online Documentation") This is layer normalization defined in ONNX as function. The overall computation can be split into two stages. The first stage is standardization, which makes the normalized elements have zero mean and unit variances. The computation required by standardization can be described by the following equations. ``` Mean = ReduceMean(X) D = Sub(X, Mean) DD = Mul(D, D) Var = ReduceMean(DD) VarEps = Add(Var, epsilon) StdDev = Sqrt(VarEps) InvStdDev = Reciprocal(StdDev) Normalized = Mul(D, InvStdDev) ``` where `normalized_axes` is `[axis, ..., rank of X - 1]`. The variables `Var` and `StdDev` stand for variance and standard deviation, respectively. The second output is `Mean` and the last one is `InvStdDev`. Depending on `stash_type` attribute, the actual computation must happen in different floating-point precision. For example, if `stash_type` is 1, this operator casts all input variables to 32-bit float, perform the computation, and finally cast `Normalized` back to the original type of `X`. The second stage then scales and shifts the outcome of the first stage using ``` NormalizedScaled = Mul(Normalized, Scale) Y = Add(NormalizedScaled, B) ``` The second stage doesn't depends on `stash_type`. All equations are in [this syntax](https://github.com/onnx/onnx/blob/main/docs/Syntax.md). The same variable (i.e., input, output, and attribute) uses the same name in the equations above and this operator's definition. Let `d[i]` indicate the i-th dimension of `X`. If `X`'s shape is `[d[0], ..., d[axis-1], d[axis], ..., d[rank-1]]`, the shape of `Mean` and `InvStdDev` is `[d[0], ..., d[axis-1], 1, ..., 1]`. `Y` and `X` have the same shape. This operator supports unidirectional broadcasting (tensors `Scale` and `B` should be unidirectional broadcastable to tensor `X`); for more details please check `Broadcasting in ONNX `_. Args: X: Tensor to be normalized. Scale: Scale tensor. B: (optional) Bias tensor. axis: The first normalization dimension. If rank(X) is r, axis' allowed range is [-r, r). Negative value means counting dimensions from the back. epsilon: The epsilon value to use to avoid division by zero. stash_type: Type of Mean and InvStdDev. This also specifies stage one's computation precision. LayerNormalizationr$r#rRr2) r5XScaleBr=rSrTr7r8s r'rVOpset17.LayerNormalizationsIR0"b9 *F 3  ! !&U 6!   r)T1_MelWeightMatrixT2_MelWeightMatrixT3_MelWeightMatrix)r.c n[SSS5n[USU5nU"URUUUUUU5SU06$)u[🌐 MelWeightMatrix(17)](https://onnx.ai/onnx/operators/onnx__MelWeightMatrix.html#melweightmatrix-17 "Online Documentation") Generate a MelWeightMatrix that can be used to re-weight a Tensor containing a linearly sampled frequency spectra (from DFT or STFT) into num_mel_bins frequency information based on the [lower_edge_hertz, upper_edge_hertz] range on the mel scale. This function defines the mel scale in terms of a frequency in hertz according to the following formula: mel(f) = 2595 * log10(1 + f/700) In the returned matrix, all the triangles (filterbanks) have a peak value of 1.0. The returned MelWeightMatrix can be used to right-multiply a spectrogram S of shape [frames, num_spectrogram_bins] of linear scale spectrum values (e.g. STFT magnitudes) to generate a "mel spectrogram" M of shape [frames, num_mel_bins]. Args: num_mel_bins: (non-differentiable) The number of bands in the mel spectrum. dft_length: (non-differentiable) The size of the original DFT. The size of the original DFT is used to infer the size of the onesided DFT, which is understood to be floor(dft_length/2) + 1, i.e. the spectrogram only contains the nonredundant DFT bins. sample_rate: (non-differentiable) Samples per second of the input signal used to create the spectrogram. Used to figure out the frequencies corresponding to each spectrogram bin, which dictates how they are mapped into the mel scale. lower_edge_hertz: (non-differentiable) Lower bound on the frequencies to be included in the mel spectrum. This corresponds to the lower edge of the lowest triangular band. upper_edge_hertz: (non-differentiable) The desired top edge of the highest frequency band. output_datatype: The data type of the output tensor. Strictly must be one of the values from DataType enum in TensorProto whose values correspond to T3. The default value is 1 = FLOAT. MelWeightMatrixr$r#r.r2) r5 num_mel_binsrC sample_ratelower_edge_hertzupper_edge_hertzr.r7r8s r'r_Opset17.MelWeightMatrixdsZ`-r26 ' 0  ! !    ,  r)T1_STFTT2_STFT)r?c h[SSS5n[USU5nU"URXaX#U5SU06$)u;[🌐 STFT(17)](https://onnx.ai/onnx/operators/onnx__STFT.html#stft-17 "Online Documentation") Computes the Short-time Fourier Transform of the signal. Args: signal: (non-differentiable) Input tensor representing a real or complex valued signal. For real input, the following shape is expected: [batch_size][signal_length][1]. For complex input, the following shape is expected: [batch_size][signal_length][2], where [batch_size][signal_length][0] represents the real component and [batch_size][signal_length][1] represents the imaginary component of the signal. frame_step: (non-differentiable) The number of samples to step between successive DFTs. window: (optional, non-differentiable) A tensor representing the window that will be slid over the signal.The window must have rank 1 with shape: [window_shape]. It's an optional value. frame_length: (optional, non-differentiable) A scalar representing the size of the DFT. It's an optional value. onesided: If onesided is 1, only values for w in [0, 1, 2, ..., floor(n_fft/2) + 1] are returned because the real-to-complex Fourier transform satisfies the conjugate symmetry, i.e., X[m, w] = X[m,w]=X[m,n_fft-w]*. Note if the input or window tensors are complex, then onesided output is not possible. Enabling onesided with real inputs performs a Real-valued fast Fourier transform (RFFT).When invoked with real or complex valued input, the default value is 1. Values can be 0 or 1. STFTr$r#r?r2)r5signal frame_stepwindow frame_lengthr?r7r8s r'rh Opset17.STFTsHTFB+ ff %  ! !&*l S   r) S_SequenceMap V_SequenceMapch[SSS5n[USU5nU"UR"XA/UQ76SU06$)u[🌐 SequenceMap(17)](https://onnx.ai/onnx/operators/onnx__SequenceMap.html#sequencemap-17 "Online Documentation") Applies a sub-graph to each sample in the input sequence(s). Inputs can be either tensors or sequences, with the exception of the first input which must be a sequence. The length of the first input sequence will determine the number of samples in the outputs. Any other sequence inputs should have the same number of samples. The number of inputs and outputs, should match the one of the subgraph. For each i-th element in the output, a sample will be extracted from the input sequence(s) at the i-th position and the sub-graph will be applied to it. The outputs will contain the outputs of the sub-graph for each sample, in the same order as in the input. This operator assumes that processing each sample is independent and could executed in parallel or in any order. Users cannot expect any specific ordering in which each subgraph is computed. Args: input_sequence: Input sequence. additional_inputs: (variadic, heterogeneous) Additional inputs to the graph body: The graph to be run for each sample in the sequence(s). It should have as many inputs and outputs as inputs and outputs to the SequenceMap function. SequenceMapr$r#bodyr2)r5input_sequencerradditional_inputsr7r8s r'rqOpset17.SequenceMap sCDM2r2 mV ,4''SARS_Z^__r))r6r*r.intr/rwreturnr+)N) rBr:rCzOptional[T2_DFT]r=rwr>rwr?rwrxr:)r6rEr.rwr/rwrxrF)r6rJr.rwr/rwrxrK)rWrOrXrOrYzOptional[T_LayerNormalization]r=rwrSfloatrTrwrxzGTuple[T_LayerNormalization, U_LayerNormalization, U_LayerNormalization])r`r[rCr[rar[rbr\rcr\r.rwrxr])NN) rirerjrfrkzOptional[T1_STFT]rlzOptional[T2_STFT]r?rwrxre)rsrnrtrorrr rxrn)3__name__ __module__ __qualname____firstlineno__r%rrrr*rr rrrrrrrrrr+__annotations__r1r:r;rArErFrHrJrKrMrOrPrVr[r\r]r_rerfrhrrrrrrnrorq__static_attributes__rvr)r'r r +s* 3UEB#(        $y  BCTU % ;> NQ  >Xx @F Xue ,F (,6 6 6 %6  6  6 6  6 p15%@"'        #i  ABST $ := MP  >OUE:M$        M9  >?PQ ! 7: JM  >##98VUT[\&+HeO&<)< -1 P .P P $P  * P  P P P  QP d!!5ueD !5xPWX$)        % . !< (< '< ( < - < - < <  < |i65'BGi.G %)*. / / / / " / ( / /  / b!M&                              ! " # $ % & ' ( ) * + , - . / 0 1 2 3 4 5 6 7 8 9 : ; < = > ? MD$`%$`*$` $`  $`r)r N)% __future__rtypingrrrrronnxr onnx.defsr typing_extensionsr #onnxscript.onnx_opset._impl.opset16r onnxscript.onnx_typesr rrrrrrrrrrrrrrronnxscript.valuesrrr rvr)r'rsJ#<< '7$(E`gE`r)